SimFuse: A Novel Fusion Simulator for RNA Sequencing (RNA-Seq) Data

Area of Science:

• Bioinformatics
• Genomics
• Computational Biology

Background:

• Accurate performance evaluation of fusion detection algorithms requires data with known gene rearrangements.
• Simulated data offers advantages over real data for estimating accuracy measures like precision and recall.
• Existing simulated RNA sequencing (RNA-Seq) datasets are limited in size, hindering systematic algorithm evaluation.

Purpose of the Study:

• To introduce SimFuse, a novel tool for generating simulated RNA-Seq fusion datasets.
• To address the limitations of existing simulated datasets for evaluating fusion detection algorithms.
• To enable rigorous validation and comparison of RNA-Seq based fusion detection tools.

Main Methods:

• SimFuse uses real sequencing data for background, mimicking real library read distributions.
• It employs a reference genome to simulate fusion-supporting reads.
• The tool generates datasets with varying numbers of fusion-supporting reads for detailed analysis.

Main Results:

• SimFuse provides enhanced control over supporting read features and simulated library size compared to existing methods.
• Enables robust estimation of performance metrics crucial for algorithm validation.
• Facilitates systematic evaluation of RNA-Seq fusion detection algorithms.

Conclusions:

• SimFuse is a valuable resource for the bioinformatics community, enabling comprehensive assessment of fusion detection tools.
• The flexibility of SimFuse allows for tailored simulations to address specific research questions.
• Improved simulation capabilities will drive advancements in the accuracy and reliability of fusion detection in genomics.